MLS Project

MLS Soccer Project

In this project I will be examining the data about MLS soccer players. It started out by finding a data set on kaggle.com by Joseph Mohr. The data consisted of players, goalkeepers, matches, and standings for the MLS starting from the beginning of the season, but the data was limited to player game stats. From there I found player salary data from the MLS Player’s Association yearly reports. That proved to be challenging as the files were yearly and in pdf form, but after downloading them to adobe and importing them as excel files I was able to clean the data and merge them to the original data. The salary data was handy, but of course I wanted more. I went searching for race data to answer the question, does race impact player wages, but I could not find race data publicly available. Instead, I found data on player nationalities from FBref, I picked 2017-2020 as the kaggle dataset only ran through 2020. The final dataset has players and goalkeepers combined with player stats, player salaries, nationality, and club.

Now that we have our datasets for players we can examine some data about them.

Descriptive Stats

Player Minutes

describe(Active_Players$MINS)

##    vars     n  mean     sd median trimmed    mad min  max range skew kurtosis
## X1    1 12747 946.5 876.76    596  855.17 769.47   1 3060  3059 0.67    -0.92
##      se
## X1 7.77

# Active_Players %>%
#   filter(Year == 2020) %>%
#   ggplot(aes(x = MINS, y = POS))+
#   geom_col(fill = 'Cornflowerblue')+
#   facet_wrap(~Club)+
#   theme_minimal()

# Weird, there is no club with the abbreviation LFC in the MLS. I looked into it and it looks like it is actually meant to be LAFC. So I will mutate the cases of LFC to LAFC.

Active_Players$Club <-replace(Active_Players$Club, Active_Players$Club == "LFC", "LAFC")
Active_Players$Club <-replace(Active_Players$Club, Active_Players$Club == "NY", "NYC")
Active_Players$Club <-replace(Active_Players$Club, Active_Players$Club == "MCF", "MTL")

# I had noticed there was a similar issue for New York City FC and Montreal CF

Active_Players %>%
  filter(Year == 2020) %>%
  ggplot(aes(x = MINS, y = POS))+
  geom_col(fill = 'Darkorchid2')+
  facet_wrap(~Club)+
  theme_minimal()+
  labs(x = "Minutes Played", y = "Position Played", title = "Minutes played by Position in 2020",
       subtitle = "Separated by Club", caption = "Data from Joseph Mohr: Major League Soccer Dataset")

To be honest, this data is not all that helpful in visualizing anything. What I would be more interested in showing would be number of players used in a season by club. To show which teams used a wide array of players compared to those that had a core group.

#First let's look at the format
str(Active_Players)

## 'data.frame':    12747 obs. of  28 variables:
##  $ Player : chr  "Roy Lassiter" "Raul Diaz Arce" "Eduardo Hurtado" "Preki" ...
##  $ Club   : chr  "TB" "DC" "LA" "KC" ...
##  $ POS    : chr  "F" "F" "F" "M" ...
##  $ GP     : int  30 28 26 32 28 26 28 31 31 26 ...
##  $ GS     : int  30 28 26 32 28 20 25 31 29 15 ...
##  $ MINS   : int  2580 2351 2323 2880 2307 1649 2233 2654 2564 1576 ...
##  $ G      : int  27 23 21 18 17 14 13 13 13 13 ...
##  $ A      : int  4 2 7 13 3 3 7 5 5 1 ...
##  $ SHTS   : int  76 100 87 140 79 54 70 58 90 49 ...
##  $ SOG    : int  49 49 56 61 44 32 48 29 41 32 ...
##  $ GWG    : int  4 4 6 3 3 6 3 4 3 3 ...
##  $ PKG/A  : chr  "3/3" "4/4" "2/3" "7/8" ...
##  $ HmG    : int  19 15 13 14 12 11 10 3 7 8 ...
##  $ RdG    : int  8 8 8 4 5 3 3 10 6 5 ...
##  $ G/90min: num  0.94 0.88 0.81 0.56 0.66 0.76 0.52 0.44 0.46 0.74 ...
##  $ SC%    : num  35.5 23 24.1 12.9 21.5 25.9 18.6 22.4 14.4 26.5 ...
##  $ GWA    : int  2 0 0 2 0 0 0 0 0 0 ...
##  $ HmA    : int  2 0 4 9 2 3 4 5 3 1 ...
##  $ RdA    : int  2 2 3 4 1 0 3 0 2 0 ...
##  $ A/90min: num  0.14 0.08 0.27 0.41 0.12 0.16 0.28 0.17 0.18 0.06 ...
##  $ FC     : int  20 32 48 26 21 27 35 36 38 27 ...
##  $ FS     : int  39 26 26 44 46 42 42 69 48 22 ...
##  $ OFF    : int  70 35 25 7 10 20 44 30 7 38 ...
##  $ YC     : int  2 6 5 3 0 4 0 2 4 1 ...
##  $ RC     : int  0 1 0 0 0 0 1 0 0 0 ...
##  $ SOG%   : num  64.5 49 64.4 43.6 55.7 ...
##  $ Year   : int  1996 1996 1996 1996 1996 1996 1996 1996 1996 1996 ...
##  $ Season : chr  "reg" "reg" "reg" "reg" ...

# Let's change the Position to factors. 
Active_Players$POS <- as.factor(Active_Players$POS)
Active_Players$Club <- as.factor(Active_Players$Club)

# Player Plot with all players but it is too much
# player_plot <- Active_Players %>%
# filter(Year == 2020) %>%
# group_by(Club, Player) %>%
# mutate(Player = reorder(Player, MINS)) %>%
# ggplot(aes(MINS, Player)) +
# geom_col(fill = "#5A9D5A")+
# theme_minimal()
# player_plot

# Active_Players %>%
#   group_by(Club) %>%
#   filter(Year == 2020) %>%
#   count(Player)%>%
#   summarise(Player_Total = sum(Player))

needs(data.table, forcats)

Active_Players %>%
  filter(Year == 2020, Club != "")%>%
  count(Club) %>%
  mutate(Club = fct_reorder(Club, desc(n))) %>%
  ggplot()+
  geom_col(aes(x = n, y = Club, fill = Club), show.legend = FALSE)+
  theme_minimal()+
  labs(x = "Number of Players", y = "")+
  scale_x_continuous(name="Number of Active Field Players", breaks=seq(0,30,3))+
  scale_fill_viridis_d()

One of the other areas of interest would be to gather data on player salaries and add it to the dataset to examine which players had the best “bang for your buck” on goals, passes, saves, etc.

There is data on salaries available, but it is a pdf document so I need to learn how to get the data off a pdf into a data structure I can gather in R.

Number of players

Active_Players %>%
  filter(Year == 2020, Club != "")%>%
  count(Club) %>%
  mutate(Club = fct_reorder(Club, desc(n))) %>%
  ggplot()+
  geom_col(aes(x = n, y = Club, fill = Club), show.legend = FALSE)+
  theme_minimal()+
  labs(x = "Number of Players", y = "")+
  scale_x_continuous(name="Number of Active Field Players", breaks=seq(0,30,3))+
  scale_fill_viridis_d()

match <- import(here("Data/MLS/matches.csv"))

This chunk has been added because I realized that I needed the goal keeper data to have an accurate display of club payments. In addition, I figured that including all players that I could would give me the most accurate display of club expenditure on player wages, even if the player did not play. So this chunk did not filter out players with zero minutes, but is limited to 2017-2020.

Data Cleaning

### Need to merge GK and Player data-sets for salary
MLS_Players17_20 <- MLS_Players %>% filter(Year == 2017 | Year == 2018 | Year == 2019 | Year == 2020)

MLS_Players17_20 <- MLS_Players17_20 %>% select(-c("Club","GWG", "PKG/A", "GWA"))

MLS_GK17_20 <- MLS_GK %>% filter(Year == 2017 | Year == 2018 | Year == 2019 | Year == 2020)

MLS_GK17_20 <- MLS_GK17_20 %>% select(-c("Club", "PKG/A", "W", "L", "T", "W%"))


Players_all_17_20 <- merge(MLS_Players17_20, MLS_GK17_20, by = c("Player", "POS", "MINS","GP", "GS", "Year", "Season"), all = TRUE)

Players_all_17_20$Player <- stri_trans_general(Players_all_17_20$Player, id = "Latin-ASCII")

Players_all_17_20$Player <- gsub("-", "", Players_all_17_20$Player)

### Reading in the 2019 player salaries
sal_19 <- read_xlsx(here("Data/Salary 19.xlsx"), skip = 1)

### Adjusting the column names
fix <- c(Base_Salary_19 = "Base Salary 19", Total_Comp_19 = "Base Guaranteed Comp 19")
sal_19 <- rename(sal_19, all_of(fix))

### Combining first and last name to create a combined player name
sal_19$Player <- apply(sal_19[, c("First Name", "Last Name")], 1,
                      function(i){ paste(na.omit(i), collapse = " ") })

### Removing accents from names as the original list to merge to does not have accents
sal_19$Player <- stri_trans_general(sal_19$Player, id = "Latin-ASCII")

### Removing hyphens between names
sal_19$Player <- gsub("-", "", sal_19$Player)

### Selecting only club, salary, and name
sal_19 <- sal_19 %>% select(c("Club", "Base_Salary_19", "Total_Comp_19", "Player"))



# I need to figure out how to get the player name to merge correctly. There is some missing data for player names that are not identical.

# 2019 Salaries are in, but let's add in 2017, 2018, and 2020
###############################################################################
### Importing and cleaning 2017 salaries
sal_17 <- read_xlsx(here("Data/Salary 17.xlsx"), skip = 1, col_names = TRUE)
### Fixing weird spacing in the name column that caused errors
fix <- c(First_Name = "First Name", Last_Name = "Last Name", Base_Salary_17 = "Base Salary", Total_Comp_17 = "Compensation")
sal_17 <- rename(sal_17, all_of(fix))

### Combining first and last name while ignoring NAs
sal_17$Player <- apply(sal_17[, c("First_Name", "Last_Name")], 1,
                      function(i){ paste(na.omit(i), collapse = " ") })

### Removing any accent marks in names to combine with existing data
sal_17$Player <- stri_trans_general(sal_17$Player, id = "Latin-ASCII")

### Remove dash from names to match with original data
sal_17$Player <- gsub("-", "", sal_17$Player)

### Selecting only club, salary, and name
sal_17 <- sal_17 %>% select(c("Club", "Base_Salary_17", "Total_Comp_17", "Player"))


###############################################################################
### Importing 2018 player salaries
sal_18 <- read_xlsx(here("Data/Salary 18.xlsx"))

### Fixing column names
fix <- c(First_Name = "First Name", Last_Name = "Last Name", Base_Salary_18 = "Salary ", Total_Comp_18 = "Compensation ")
sal_18 <- rename(sal_18, all_of(fix))

sal_18$Player <- apply(sal_18[, c("First_Name", "Last_Name")], 1,
                      function(i){ paste(na.omit(i), collapse = " ") })

### Removing any accent marks in names to combine with existing data
sal_18$Player <- stri_trans_general(sal_18$Player, id = "Latin-ASCII")

### Remove dash from names to match original data
sal_18$Player <- gsub("-", "", sal_18$Player)

### Selecting only club, salary, and name
sal_18 <- sal_18 %>% select(c("Club", "Base_Salary_18", "Total_Comp_18", "Player"))

### Changing the salary to numeric from character
sal_18$Base_Salary_18 <- as.numeric(sal_18$Base_Salary_18)

sal_18$Total_Comp_18 <- as.numeric(sal_18$Total_Comp_18)



###############################################################################
### Importing 2020 Player salaries
sal_20 <- read_xlsx(here("Data/Salary 20.xlsx"))

fix <- c(Base_Salary_20 = "Base Salary 20", Total_Comp_20 = "Base Guaranteed Comp 20")
sal_20 <- rename(sal_20, all_of(fix))

sal_20$Player <- apply(sal_20[, c("First Name", "Last Name")], 1,
                      function(i){ paste(na.omit(i), collapse = " ") })

sal_20$Player <- stri_trans_general(sal_20$Player, id = "Latin-ASCII")

sal_20 <- sal_20 %>% select(-contains('...6'))

### Remove dash from names to match original data
sal_20$Player <- gsub("-", "", sal_20$Player)

### Selecting only club, salary, and name
sal_20 <- sal_20 %>% select(c("Club", "Base_Salary_20", "Total_Comp_20", "Player"))




###############################################################################
# I am sure there is a better way but I don't know how
### Combining 2017 and 2018 data
Salaries <- merge(sal_17, sal_18, by = "Player", all = TRUE)

### Fixing the club column names
fix <- c(Club_17 = "Club.x", Club_18 = "Club.y")
Salaries <- rename(Salaries, all_of(fix))

### Adding in the 2019 salaries
Salaries <- merge(Salaries, sal_19, by = "Player", all = TRUE)

### Adding in the 2020 salaries
Salaries <- merge(Salaries, sal_20, by = "Player", all = TRUE)

### Fixing the club column names again
fix <- c(Club_19 = "Club.x", Club_20 = "Club.y")
Salaries <- rename(Salaries, all_of(fix))

### Selecting just the club information
Player_Club <- Salaries %>% select("Player", "Club_17", "Club_18", "Club_19", "Club_20")

### Making the club data longer
Player_Club_long <- Player_Club %>% pivot_longer(cols  = -Player, names_to = c("Club", "Year"), 
    names_sep = "\\.")

### Separating the club column by year
Player_Club_long <- Player_Club_long %>%
  separate(Club, c('Club', 'Year'), sep = "(.*)_()")

### Adding 20 infront of the years
Player_Club_long$Year <- sub("^", "20", Player_Club_long$Year)

### Making year numeric
Player_Club_long$Year <- as.numeric(Player_Club_long$Year)

### selecting only a few columns
Player_Club_long <- Player_Club_long %>% select("Player", "Year", "value")

### Changing the column name to club
Player_Club_long <- rename(Player_Club_long, Club = "value")

### Removing players with NA values for club meaning they did not play in the league that year
Player_Club_long <- Player_Club_long %>% filter(Club != "NA")

#################################################################

### Salaries needs to be in long format in order to be merged 
### Selecting just the salary data
Salaries_long <- Salaries %>% select("Player", "Base_Salary_17", "Total_Comp_17", "Base_Salary_18", "Total_Comp_18", "Base_Salary_19", "Total_Comp_19", "Base_Salary_20", "Total_Comp_20")

### pivoting longer by salary type
Salaries_long <- Salaries_long %>% pivot_longer(cols  = -Player, names_to = c("Salary", "Year"), 
    names_sep = "\\.")

### separating column by year
Salaries_long <- Salaries_long %>%
  separate(Salary, c('Salary', 'Year'), sep = "()_(.*)_()")

### adding 20 in front of years to make it match with original data
Salaries_long$Year <- sub("^", "20", Salaries_long$Year)

### Making year numeric
Salaries_long$Year <- as.numeric(Salaries_long$Year)


  
###############################################################################

### Combining the salaries into the active players list by Player name
Players_all_17_20 <- left_join(Players_all_17_20, Salaries_long, by = c("Player", "Year"))


### Simplifying the position factor to three variables
Players_all_17_20$POS <- fct_collapse(Players_all_17_20$POS, M = c("M","M-F", "M-D"), D = c("D","D-M"), F = c("F-M", "F"), Gk = "GK")

### Finally I need to rename some columns
fix <- c(Games = "GP", Starts = "GS", Goals = "G", Assists = "A", Shots = "SHTS.x", `Shots On Goal` = "SOG", `Home Goals` = "HmG", `Away Goals` = "RdG", `Goals/90` = "G/90min", `Home Assists` = "HmA", `Away Assists` = "RdA", `Scoring Percent` = "SC%", `Assists/90` = "A/90min", Fouls= "FC", `Fouls Suffered` = "FS", Offsides = "OFF", `Yellow Cards` = "YC", `Red Cards` = "RC", `Shots Faced` = "SHTS.y", Saves = "SV", `Goals Against` = "GA", `Goals Against Avg` = "GAA", Shoutouts = "ShO", `Save Pct` = "Sv%")
Players_all_17_20 <- rename(Players_all_17_20, all_of(fix))


# Now, I need to add in the correct club information because in this review I realized that the club info in the original dataset was incorrect. To do this I am going to remove the Club column and merge in the player club dataset I took from the player salaries.

Players_all_17_20 <- left_join(Players_all_17_20, Player_Club_long, by = c("Player", "Year"))


###############################################################################

# I realized I need to pivot the salary data into a wider format

Players_all_17_20 <- Players_all_17_20 %>%
  pivot_wider(
    names_from = Salary,
    values_from = value)

### For some reason there is a weird empty column now. I am sure I probably pivoted incorrectly, but I am just going to delete the extra column because that seems easier since I have the data in the right place

Players_all_17_20 <- Players_all_17_20 %>% select(-c("NA"))

fix <- c(`Base Salary` = "Base", `Total Comp` = "Total")
Players_all_17_20 <- rename(Players_all_17_20, all_of(fix))

# Much cleaner!


#####################################################################

# I am still deciding if I need/want to include percent change in salary

### Adding in salary change amount
Players_all_17_20$Difference <- (Players_all_17_20$`Total Comp`-Players_all_17_20$`Base Salary`)

### Adding salary percent change by season
Players_all_17_20$Pct_Difference <- round(((Players_all_17_20$Difference)/(Players_all_17_20$`Total Comp`))*100,digits=2)

### Creating a subset of regular season games
Players_all_17_20_Reg <- Players_all_17_20 %>%
  filter(Season == "reg")

Just when I think I have solved it, new problems crop up. It seems like there is a mistake in the active players list so the clubs are inaccurate. I can either pivot the clubs from the salaries data and join it into the main dataset or try to figure out why the clubs are not showing correctly in the active players data.

I also want to calculate percent change, but I might need to do that before pivoting the data. Just kidding! After pivoting wider, I can now do it easily in the cleaned data.

Goals and Assists

Players_all_17_20 <- Players_all_17_20 %>%
  group_by(Player) %>%
  add_count(Player)

options(scipen = 999)



sal_19_plot <- Players_all_17_20 %>%
  filter(Year == 2019 & Season == "reg") %>%
  ggplot(aes(x = `Goals/90`, y = `Total Comp`), label = Player) +
  geom_point(aes(color = POS, text = Player), show.legend = F) +
  coord_flip() +
  facet_wrap(~POS)+
  theme_minimal() +
  labs(title = "Goals per 90 by Total Compensation") +
  scale_x_log10(n.breaks = 8) +
  xlim(0,4)

sal_19_plot + stat_cor(aes(x = `Goals/90`, y = `Total Comp`), data = Players_all_17_20, label.y = 10000, label.x = 4.2,  p.accuracy = 0.001)

ggplotly(sal_19_plot, tooltip = c("Player", "Total Comp", "Goals/90"))

sal_19_plot2 <- Players_all_17_20 %>%
  filter(Year == 2019 & Season == "reg") %>%
  ggplot(aes(x = `Assists/90`, y = `Total Comp`), label = Player) +
  geom_point(aes(color = POS, text = Player), show.legend = F) +
  coord_flip() +
  facet_wrap(~POS)+
  theme_minimal() +
  labs(title = "Assists per 90 by total compensation") +
  scale_x_log10(n.breaks = 8) +
  xlim(0,1.5)

ggplotly(sal_19_plot2, tooltip = c("Player", "Total Comp", "Assists/90", "MINS"))

sal_19_plot3 <- Players_all_17_20 %>%
  filter(Year == 2019 & Season == "reg") %>%
  select(Player, Goals, `Total Comp`, POS, MINS) %>%
  ggplot(aes(x = Goals, y = `Total Comp`), label = Player) +
  geom_point(aes(color = POS, text = Player), show.legend = F) +
  coord_flip() +
  facet_wrap(~POS)+
  theme_minimal() +
  labs(title = "Goals by Total Compensation") +
  scale_x_log10(n.breaks = 8)

ggplotly(sal_19_plot3, tooltip = c("Player", "Total Comp", "Goals"))

Salary Data

Finally, we can look at some player salary data

I mean I can see the players but it is too crowded to get much from it. I am going to look at Minnesota United since they are the team I support.

sal_plot <- Players_all_17_20_Reg %>%
  filter(Club == "Minnesota United") %>%
  group_by(Player, Year) %>%
  ggplot(aes(x = Year, y = `Total Comp`), label = Player) +
  geom_line(aes(color = POS, text = Player), show.legend = F) +
  theme_minimal() +
  scale_y_log10(n.breaks = 8) +
  labs(title = "Minnesota United Player wages by year")

ggplotly(sal_plot, tooltip = c("Player", "Total Comp"))

Still not that helpful

Club level salaries

Players_all_17_20_Reg %>%
  filter(Club != "Retired" & Club != "Major League Soccer") %>%
  group_by(Year, Club) %>%
  summarize(Club_avg = mean(`Total Comp`)) %>%
  ggplot(aes(x = Club_avg, y = reorder(Club, Club_avg))) +
  geom_col(aes(fill = Club), position = "dodge", show.legend = FALSE) +
  theme_minimal() +
  facet_wrap(~Year) +
  scale_fill_viridis_d() +
  coord_flip() +
  labs(x = "Average Total Compensation", y = "", title = "Average Total Compensation by Club (By Year: 2017-2020)") +
  theme(axis.text.x = element_text(angle = 45, vjust = 1.1, hjust = 1))

Players_all_17_20_Reg %>%
  filter(Club != "Retired" & Club != "Major League Soccer") %>%
  group_by(Year, Club) %>%
  summarize(Club_avg = sum(`Total Comp`)) %>%
  ggplot(aes(x = Club_avg, y = reorder(Club, Club_avg))) +
  geom_col(aes(fill = Club), position = "dodge", show.legend = FALSE) +
  theme_minimal() +
  facet_wrap(~Year) +
  scale_fill_viridis_d(option = "magma") +
  coord_flip() +
  labs(x = "Sum Total Compensation", y = "", title = "Total Compensation of Player Wages by Club (By Year: 2017-2020)") +
  theme(axis.text.x = element_text(angle = 45, vjust = 1.1, hjust = 1))

Players_all_17_20_Reg %>%
  filter(Club != "Retired" & Club != "Major League Soccer") %>%
  group_by(Club) %>%
  summarize(Club_avg = sum(`Total Comp`)) %>%
  ggplot(aes(x = Club_avg, y = reorder(Club, Club_avg))) +
  geom_col(aes(fill = Club), position = "dodge", show.legend = FALSE) +
  theme_minimal() +
  scale_fill_viridis_d() +
  coord_flip() +
  labs(x = "Average Wages", y = "", "Total ", title = "Average wages across years by Club") +
  theme(axis.text.x = element_text(angle = 45, vjust = 1.1, hjust = 1))

I know this doesn’t look like much but it took so long to get this data cleaned and into R.

Now I think I should probably add in the goal keeper data so that I can see the real club expenditure.

In addition, I probably should keep all players, even those that didn’t play.

Update

I think the player data has been cleaned now.

If I want to answer the question about does player data x have a correlation with club stat y I will need to read in match data to use, but for now, I can at least compare players across 2017-2018 and club expenditure as well.

Foul Data

sal_plot <- Players_all_17_20_Reg %>%
  filter(Year == 2017 & POS != "GK") %>%
  group_by("POS") %>%
  ggplot() +
  geom_point(aes(x = `Total Comp`, y = `Fouls Suffered`, color = POS, text = Player), show.legend = T) +
  theme_fivethirtyeight()+
  labs(title = "Fouls Suffered by Salary", subtitle = "Position, not salary determine fouls suffered")

ggplotly(sal_plot, tooltip = c("Player", "Fouls Suffered"))

sal_plot <- Players_all_17_20_Reg %>%
  filter(Year == 2017 & POS != "GK") %>%
  group_by("POS") %>%
  ggplot() +
  geom_point(aes(x = `Total Comp`, y = `Fouls`, color = POS, text = Player), show.legend = T) +
  theme_fivethirtyeight()+
  labs(title = "Fouls by Salary", subtitle = "Position, not salary determine fouls")

ggplotly(sal_plot, tooltip = c("Player", "Fouls"))

sal_plot <- Players_all_17_20_Reg %>%
  filter(Year == 2017 & POS != "GK") %>%
  group_by("POS") %>%
  ggplot() +
  geom_point(aes(x = `Fouls Suffered`, y = `Fouls`, color = POS, text = Player), show.legend = T, position = "jitter") +
  theme_minimal()+
  labs(title = "Fouls by Fouls Suffered", x = "Fouls Suffered", y = "Fouls")

ggplotly(sal_plot, tooltip = c("Player", "Fouls", "Fouls Suffered"))

I realized that I would like to add in race/ethnicity data but in the process I found out that I could not find that data. However, I was able to find data on player nationality based on playing youth or senior level international games and/or player stated nationality. I recognize it is not the same as race data as a player from the US could be black, white, latine, or other races, but nationality is as close as I can get. Other countries, such as Colombia or Mexico are more homogeneous. I will not make any claims of race though to be cautious.

Nationality Data

require(countrycode)

## Loading required package: countrycode

### Data from Fbref on 2020 player data and nationality
# Importing
Nation_20 <- import(here("Data/2020_player_nation_sal.csv"))

# Removing the wages data because I already have this data and there are discrepancies
Nation_20 <- Nation_20 %>% select(-c("Weekly Wages", "Annual Wages"))

# From homework 4, getting the country codes
country_codes <- countrycode::codelist %>% 
  filter(ecb != 'NA') %>%
  select(country_name = country.name.en, country = ecb) 

# making the country codes lower case because I need to match with Nation_20
country_codes$country <- str_to_lower(country_codes$country)

# Merging together to get country names
Nation_20 <- merge(Nation_20, country_codes, by =  "country")

# Remvoing accent marks from names to make merging easier
Nation_20$Player <- stri_trans_general(Nation_20$Player, id = "Latin-ASCII")

### Remove dash from names to match original data
Nation_20$Player <- gsub("-", "", Nation_20$Player)


### Merging but I need to figure out how to merge year so that it does not delete the 2017 data.
Players_all_17_20_Reg_Nation <- merge(Players_all_17_20_Reg, Nation_20, by = c("Player", "Year"), all = TRUE)

### Removing duplicate column and empty column
Players_all_17_20_Reg_Nation <- Players_all_17_20_Reg_Nation %>% select(-c("V9"))

This dataset is getting to be a little overwhelming, but I think it has all the information needed now.

## Creating a column by country count
Players_all_17_20_Reg_Nation <- Players_all_17_20_Reg_Nation %>%
  group_by(country, Year) %>%
  add_count(country)

## average compensation
Players_all_17_20_Reg_Nation <- Players_all_17_20_Reg_Nation %>%
  group_by(country) %>%
  mutate(country_comp = mean(`Total Comp`, na.rm = TRUE))

## average minutes
Players_all_17_20_Reg_Nation <- Players_all_17_20_Reg_Nation %>%
  group_by(country) %>%
  mutate(country_mins = mean(MINS, na.rm = TRUE))

## Plot of player average minutes and average comp by country
nation_player_plot <- Players_all_17_20_Reg_Nation %>%
  group_by(Player) %>%
  ggplot(aes(x = country_mins, y = country_comp)) + 
  geom_point(aes(size = n), colour = ifelse(Players_all_17_20_Reg_Nation$n > 20,"darkorchid","gray"), alpha = 0.5) +
  geom_text(data = Players_all_17_20_Reg_Nation |> dplyr::filter(n > 20),
            aes(label = country_name), check_overlap = TRUE, nudge_y = 0.1, nudge_x = -0.12) +
  scale_size_binned(name = "Players per Country", breaks = c(5,25,50)) +
  theme_minimal() +
  labs(x = "Average Minutes played per season by country", 
       y = "Average Total compensation by country") +
  theme(legend.position = 'bottom') +
  labs(title = "Higher paid players tend to play more", subtitle = "Average Player minutes by Average Total Compensation with player nationality")

ggplotly(nation_player_plot)

The United States, Argentina, Brazil, and Canada all have the most players in the MLS. The 5th group is a combination of players with no nationality data.

Average Wages by Country

Players_all_17_20_Reg_Nation %>%
  filter(Club != "Retired" & Club != "Major League Soccer" & country_name != "NA") %>%
  group_by(country_name) %>%
  ggplot(aes(x = country_comp, y = reorder(country_name, country_comp))) +
  geom_col(aes(fill = country_name), position = "dodge", show.legend = FALSE) +
  theme_minimal() +
  scale_fill_viridis_d() +
  coord_flip() +
  labs(x = "Average Wages", y = "", title = "Average wages by country") +
  theme(axis.text.x = element_text(angle = 45, vjust = 1.1, hjust = 1))

Players_all_17_20_Reg_Nation %>%
  filter(Club != "Retired" & Club != "Major League Soccer" & country_name != "NA" & n >= 15) %>%
  group_by(country_name) %>%
  ggplot(aes(x = country_comp, y = reorder(country_name, country_comp))) +
  geom_col(aes(fill = country_name), position = "dodge", show.legend = FALSE) +
  theme_minimal() +
  scale_fill_viridis_d() +
  coord_flip() +
  labs(x = "Average Wages", y = "", title = "Average wages by country (With at least 15 players in the league)") +
  theme(axis.text.x = element_text(angle = 45, vjust = 1.1, hjust = 1))

#install.packages("flexdashboard")
library(flexdashboard)